Biology subjects

Zhang, B.

Publications and source records attributed to Zhang, B..

At least 415 records · Page 23Linked to original sources

bioRxiv2022

Methamphetamine induced regional-specific transcriptomic and epigenetic changes in the rat brain

BackgroundMethamphetamine (METH) is a highly addictive central nervous system stimulant. Chronic use of METH is associated with multiple neurological and psychiatric disorders. An overdose of METH can cause brain damage and even death. Mounting evidence indicates that epigenetic changes and functional impairment in the brain occur due to addictive drug exposures. However, the responses of different brain regions to a METH overdose remain unclear. ResultsWe investigated the transcriptomic and epigenetic responses to a METH overdose in four regions of the rat brain, including the nucleus accumbens, dentate gyrus, Ammons horn, and subventricular zone. We found that 24 hours after METH overdose, 15.6% of genes showed changes in expression and 27.6% of open chromatin regions exhibited altered chromatin accessibility in all four rat brain regions. Interestingly, only a few of those differentially expressed genes and differentially accessible regions were affected simultaneously. Among four rat brain regions analyzed, 149 transcription factors and 31 epigenetic factors were significantly affected by METH overdose. METH overdose also resulted in opposite-direction changes in regulation patterns of both gene and chromatin accessibility between the dentate gyrus and Ammons horn. Approximately 70% of chromatin-accessible regions with METH-induced alterations in the rat brain are conserved at the sequence level in the human genome, and they are highly enriched in neurological processes. Many of these conserved regions are active brain-specific enhancers and harbor SNPs associated with human neurological functions and diseases. ConclusionOur results indicate strong region-specific transcriptomic and epigenetic responses to a METH overdose in distinct rat brain regions. We describe the conservation of region-specific gene regulatory networks associated with METH overdose. Overall, our study provides clues toward a better understanding of the molecular responses to METH overdose in the human brain.

bioinformatics↗

bioRxiv2022

A Caenorhabditis elegans nck-1 and filamentous actin-regulating protein pathway mediates a key cellular defense against bacterial pore-forming proteins

Pore-forming proteins (PFPs) comprise the largest single class of bacterial protein virulence factors and are expressed by many human and animal bacterial pathogens. Cells that are attacked by these virulence factors activate epithelial intrinsic cellular defenses (or INCEDs) to prevent the attendant cellular damage, cellular dysfunction, osmotic lysis, and organismal death. Several conserved PFP INCEDs have been identified using the nematode Caenorhabditis elegans and the nematicidal PFP Cry5B, including mitogen-activated protein kinase (MAPK) signaling pathways. Here we demonstrate that the gene nck-1, which has homologs from Drosophila to humans and links cell signaling with localized F-actin polymerization, is required for INCED against small-pore PFPs in C. elegans. Reduction/loss of nck-1 function results in C. elegans hypersensitivity to PFP attack, a hallmark of a gene required for INCEDs against PFPs. This requirement for nck-1-mediated INCED functions cell-autonomously in the intestine and is specific to PFPs but not other tested stresses. Genetic interaction experiments indicate that nck-1-mediated INCED against PFP attack is independent of the major MAPK PFP INCED pathways. Proteomics and cell biological and genetic studies further indicate that nck-1 functions with F-actin cytoskeleton modifying genes like arp2/3, erm-1, and dbn-1 and that nck-1/arp2/3 promote pore repair at the membrane surface and protect against PFP attack independent of p38 MAPK. Consistent with these findings, PFP attack causes significant changes in the amount of actin cytoskeletal proteins and in total amounts of F-actin in the target tissue, the intestine. nck-1 mutant animals appear to have lower F-actin levels than wild-type C. elegans. Studies on nck-1 and other F-actin regulating proteins have uncovered a new and important role of this pathway and the actin cytoskeleton in PFP INCED and protecting an intestinal epithelium in vivo against PFP attack. Author SummaryThe mechanism of action for a significant number of bacterial protein toxins is the formation of pores in the membrane of target cells. Host cells contain programmed defenses against such attacks. Here we use the model system of the roundworm Caenorhabditis elegans and crystal proteins produced by Bacillus thuringiensis to demonstrate a new defense pathway mediated by the activity of the NCK-1 protein. Consistent with its known function in mammalian cells, the NCK-1-mediated defense involves various actin-interacting proteins and affects the kinetics of pore repair. This pathway is novel in its independence from MAPK signaling. Furthermore, the NCK-1 activity is found to be generally needed for defense against multiple pore-forming proteins and yet is specific to defense against such proteins but not against other environmental stressors.

microbiology↗

bioRxiv2022

Development and application of an uncapped mRNA platform

A novel uncapped mRNA platform was developed. Five lipid nanoparticle (LNP)-encapsulated mRNA constructs were made to evaluate several aspects of our platform, including transfection efficiency and durability in vitro and in vivo and the activation of humoral and cellular immunity in several animal models. The constructs were eGFP-mRNA-LNP (for enhanced green fluorescence mRNA), Fluc-mRNA-LNP (for firefly luciferase mRNA), S{delta}T-mRNA-LNP (for Delta strain SARS-CoV-2 spike protein trimer mRNA), gDED-mRNA-LNP (for truncated glycoprotein D mRNA coding ectodomain from herpes simplex virus type 2 (HSV2)) and gDFR-mRNA-LNP (for truncated HSV2 glycoprotein D mRNA coding amino acids 1[~]400). Quantifiable target protein expression was achieved in vitro and in vivo with eGFP-and Fluc-mRNA-LNP. S{delta}T-mRNA-LNP, gDED-mRNA-LNP and gDFR-mRNA-LNP induced both humoral and cellular immune responses comparable to those obtained by previously reported capped mRNA-LNP constructs. Notably, S{delta}T-mRNA-LNP elicited neutralizing antibodies in hamsters against the Omicron and Delta strains. Additionally, gDED-mRNA-LNP and gDFR-mRNA-LNP induced potent neutralizing antibodies in rabbits and mice. The mRNA constructs with uridine triphosphate (UTP) outperformed those with N1-methylpseudouridine triphosphate (N1m{psi}TP) in the induction of antibodies via S{delta}T-mRNA-LNP. Our uncapped, process-simplified, and economical mRNA platform may have broad utility in vaccines and protein replacement drugs.

bioengineering↗

bioRxiv2022

Elemental diet enriched in amino acids alters gut microbial community and prevents colonic mucus degradation in mice with colitis

The role of dietary amino acids or intact proteins in the progression of colitis remains controversial, and the mechanism involving gut microbes is unclear. Here, we investigated the effects of an elemental diet (ED) enriched in amino acids and a polymeric diet enriched in intact protein on the pathogenesis of DSS-induced colitis in mice. Our results showed that ED induced remission of colitis in mice. Notably, ED treatment reduced the abundance of the mucolytic bacteria Akkermansia and Bacteroides, which was attributed to the decreased colonic protein fermentation. Consistently, the activities of mucolytic enzymes were decreased, leading to the protection of mucus layer degradation and microbial invasion. The fecal microbiota transplantation of ED-fed mice reshaped microbial ecology and alleviated intestinal inflammation in recipient mice. ED failed to induce remission of colitis in pseudo-germ-free mice. Together, we convincingly demonstrated the critical role of gut microbiota in the prevention of ED on colitis. ImportanceThe prevalence of inflammatory bowel disease is rapidly increasing and has become a global burden. Several specific amino acids have been shown to benefit mucosal healing and colitis remission. However, the role of amino acids or intact proteins in diets and enteral nutrition formulas is controversial, and the mechanisms involving gut microbes remain unclear. In this study, we investigated the effects of an elemental diet (ED) enriched in amino acids and a polymeric diet enriched in intact protein on the pathogenesis of colitis in mice. The underlying mechanisms were explored by utilizing fecal microbiota transplantation and pseudo-germ-free mice. ED treatment reduced the abundance of mucolytic bacteria, thereby protecting the mucus layer from microbial invasion and degradation. For the first time, we convincingly demonstrated the critical role of gut microbiota in the effects of ED. This study may provide new insights into the gut microbiota-diet interaction in human health.

microbiology↗

bioRxiv2022

A common mechanism of temperature-sensing in thermoTRP channels

Detecting temperature is crucial for the survival of living organisms. Though the thermo transient receptor potential (thermoTRP) channels, such as TRPV1 or TRPM8, have been identified as prototypic heat or cold sensors, respectively, how they detect temperature remains elusive. Here we first identified groups of clustered residues in these channels that undergo burial/exposure conformational rearrangements during temperature activation by analyzing available protein structures or hydroxyl radical footprinting-mass spectroscopy (HRF-MS). By systematically perturbing water-protein interactions at these residues, we found that the temperature sensitivity in these channels were modulated in accordance with the sidechain hydrophobicity. The changes in energy associated with changes in water-protein interactions were sufficient for thermo activation. Therefore, our study has established that the water-protein interactions as a common mechanism underlying temperature sensing in TRPM8 and TRPV1.

biophysics↗

bioRxiv2022

Neuroinflammation in neuronopathic Gaucher disease: Role of microglia and NK cells

BackgroundNeuronopathic Gaucher Disease (nGD) is a rare neurodegenerative disorder caused by biallelic mutations in Gba, and buildup of glycosphingolipids in lysosomes. Neuronal injury and cell death are prominent pathological features, however the role of Gba in individual cell types and involvement of microglia, blood derived macrophages and immune infiltrates in nGD pathology remains enigmatic. MethodsHere, using single cell resolution of mouse nGD brains, we found induction of neuroinflammation pathways involving microglia, NK cells, astrocytes, and neurons. ResultsTargeted rescue of Gba in microglia and in neurons, respectively in Gba deficient, nGD mice reversed the buildup of glucosylceramide (GlcCer) and glucosylsphingosine (GlcSph), concomitant with amelioration of neuroinflammation, reduced level of serum neurofilament light chain (Nf-L) and improved survival. The levels of bioactive lipid, GlcSph was strongly correlated with serum Nf-L and ApoE in nGD mouse models as well as GD patients. Gba rescue in microglia/macrophage compartment prolonged survival, that was further enhanced upon treatment with brain permeant inhibitor of glucosylceramide synthase, effects mediated via improved glycosphingolipid homeostasis and reversal of neuroinflammation involving activation of microglia, brain macrophages and NK cells. ConclusionsTogether, our study delineates individual cellular effects of Gba deficiency in nGD brains, highlighting the central role of neuroinflammation driven by microglia activation and the role of brain permeant small molecule glucosylceramide inhibitor in reversing complex multidimensional pathophysiology of nGD. Our findings advance disease biology whilst identifying compelling biomarkers of nGD to improve patient management, enrich clinical trials and illuminate therapeutic targets. FundingResearch grant from Sanofi Genzyme; other support includes R01NS110354.Yale Liver Center P30DK034989, pilot project grant.

immunology↗

bioRxiv2022

Phase separation and correlated motions in motorized genome

The human genome is arranged in the cell nucleus non-randomly, and phase separation has been proposed as an important driving force for genome organization. However, the cell nucleus is an active system, and the contribution of non-equilibrium activities to phase separation and genome structure and dynamics remains to be explored. We simulated the genome using an energy function parameterized with chromosome conformation capture (Hi-C) data with the presence of active, nondirectional forces that break the detailed balance. We found that active forces that may arise from transcription and chromatin remodeling can dramatically impact the spatial localization of heterochromatin. When applied to euchromatin, active forces can drive heterochromatin to the nuclear envelope and compete with passive interactions among heterochromatin that tend to pull them in opposite directions. Furthermore, active forces induce long-range spatial correlations among genomic loci beyond single chromosome territories. We further showed that the impact of active forces could be understood from the effective temperature defined as the fluctuation-dissipation ratio. Our study suggests that non-equilibrium activities can significantly impact genome structure and dynamics, producing unexpected collective phenomena.

biophysics↗

bioRxiv2022

Reconstruction of the cell pseudo-space from single-cell RNA sequencing data with scSpace

Tissues are highly complicated with spatial heterogeneity in gene expression. However, the cutting-edge single-cell RNA-seq technology eliminates the spatial information of individual cells, which contributes to the characterization of cell identities. Herein, we propose single-cell spatial position associated co-embeddings (scSpace), an integrative algorithm to distinguish spatially variable cell subclusters by reconstructing cells onto a pseudo-space with spatial transcriptome references (Visium, STARmap, Slide-seq, etc.). We demonstrated that scSpace can define biologically meaningful cell subpopulations neglected by single-cell RNA-seq or spatially resolved transcriptomics. The use of scSpace to uncover the spatial association within single-cell data, reproduced, the hierarchical distribution of cells in the brain cortex and liver lobules, and the regional variation of cells in heart ventricles and the intestinal villus. scSpace identified cell subclusters in intratelencephalic neurons, which were confirmed by their biomarkers. The application of scSpace in melanoma and Covid-19 exhibited a broad prospect in the discovery of spatial therapeutic markers.

bioinformatics↗

bioRxiv2022

Biological age is increased by stress and restored upon recovery

Aging is classically conceptualized as an ever-increasing trajectory of damage accumulation and loss of function, leading to increases in morbidity and mortality. However, recent in vitro studies have raised the possibility of age reversal. Here, we report that biological age is fluid and exhibits rapid changes in both directions. By applying advanced epigenetic aging clocks, we find that the biological age of young mice is increased by heterochronic parabiosis and restored following surgical detachment of animals. We also identify transient changes in biological age during major surgery, pregnancy, and severe COVID-19 in humans and/or mice. Together, these data show that biological age undergoes a rapid increase in response to diverse forms of stress, which is reversed following recovery from stress. Our study uncovers a new layer of aging dynamics that should be considered in future studies. Elevation of biological age by stress may be a quantifiable and actionable target for future interventions.

systems biology↗

bioRxiv2022

FoxP3-mediated blockage of ryanodine receptor 2 is the molecular basis for the contact-based suppression by regulatory T cells

The suppression mechanism of regulatory T cells is an intensely investigated topic. As our focus has shifted towards a model centered on indirect inhibition of dendritic cells, a universally applicable effector mechanism controlled by FoxP3 expression has not been found. Here, we report that FoxP3 blocks the transcription of ER Ca2+-release channel ryanodine receptor 2. Reduced RyR2 shuts down basal Ca2+ oscillation in Tregs, which reduces m-Calpain activities that is needed for T cells to disengage from DCs, suggesting a persistent blockage of DC antigen presentation. RyR2 deficiency renders the CD4+ T cell pool to become immune suppressive, and behave in the same manner as FoxP3+ Tregs in viral infection, asthma, hypersensitivity, colitis and tumor development. In the absence of FoxP3, RyR2-deficient CD4+ T cells rescue the systemic autoimmunity associated with Scurfy mice. Therefore, FoxP3-mediated Ca2+ signaling inhibition may be a central effector mechanism of Treg immune suppression. One Sentence SummaryCalcium channel RyR2 dictates Treg adhesion-based suppression

immunology↗

bioRxiv2022

Learning to suppress likely distractor locations in visual search is driven by the local distractor frequency

Salient but task-irrelevant distractors interfere less with visual search when they appear in a display region where distractors have appeared more frequently in the past ( distractor-location probability cueing). This effect could reflect the (re-)distribution of a global, limited attentional inhibition resource. Accordingly, changing the frequency of distractor appearance in one display region should also affect the magnitude of interference generated by distractors in a different region. Alternatively, distractor-location learning may reflect a local response (e.g., habituation) to distractors occurring at a particular location. In this case, the local distractor frequency in one display region should not affect distractor interference in a different region. To decide between these alternatives, we conducted three experiments in which participants searched for an orientation-defined target while ignoring a more salient orientation distractor that occurred more often in one vs. another display region. Experiment 1 varied the ratio of distractors appearing in the frequent vs. rare regions (60/40-90/10), with a fixed global distractor frequency. The results revealed the cueing effect to increase with increasing probability ratio. In Experiments 2 and 3, one ( test) region was assigned the same local distractor frequency as in one of the conditions of Experiment 1, but a different frequency in the other region - dissociating local from global distractor frequency. Together, the three experiments showed that distractor interference in the test region was not significantly influenced by the frequency in the other region, consistent with purely local learning. We discuss the implications for theories of statistical distractor-location learning. Public Significance StatementWe are frequently distracted by salient visual stimuli which are irrelevant to the task at hand. Previous studies have shown that knowledge of the location(s) where a distractor is most likely to occur helps the observer to mitigate distraction. In this study we compared different theories of how the frequency and spatial distribution of distractor occurrence in different locations could influence the ability to avoid distraction. The results favored a local learning account: the ability to avoid distraction by distractors occuring in a particular spatial region is primarily influenced by how often distractors have occurred in that region.

neuroscience↗

bioRxiv2022

INPP5D limits plaque formation and glial reactivity in the APP/PS1 mouse model of Alzheimer's disease

The dual specificity lipid/protein phosphatase SHIP1 (encoded by the INPP5D gene) is enriched in myeloid cells. Single nucleotide polymorphisms (SNPs) in INPP5D coding and non-coding regions impact risk for developing late onset sporadic Alzheimers disease (LOAD). We present pathological analyses with spatial transcriptomics of mice with tamoxifen-sensitive microglial knockdown of Inpp5d and show exacerbated plaque pathology, plaque-associated microglial density, and altered gene expression around plaques, suggesting novel markers for plaque-associated reactive microglia.

neuroscience↗

bioRxiv2022

A human embryonic limb cell atlas resolved in space and time

Human limbs emerge during the fourth post-conception week as mesenchymal buds which develop into fully-formed limbs over the subsequent months. Limb development is orchestrated by numerous temporally and spatially restricted gene expression programmes, making congenital alterations in phenotype common. Decades of work with model organisms has outlined the fundamental processes underlying vertebrate limb development, but an in-depth characterisation of this process in humans has yet to be performed. Here we detail the development of the human embryonic limb across space and time, using both single-cell and spatial transcriptomics. We demonstrate extensive diversification of cells, progressing from a restricted number of multipotent progenitors to myriad mature cell states, and identify several novel cell populations, including neural fibroblasts and multiple distinct mesenchymal states. We uncover two waves of human muscle development, each characterised by different cell states regulated by separate gene expression programmes. We identify musculin (MSC) as a key transcriptional repressor maintaining muscle stem cell identity and validate this by performing MSC knock down in human embryonic myoblasts, which results in significant upregulation of late myogenic genes. Through integration of multiple anatomically continuous spatial transcriptomic samples, we spatially map single-cell clusters across a sagittal section of a whole fetal hindlimb. We reveal a clear anatomical segregation between genes linked to brachydactyly and polysyndactyly, and uncover transcriptionally and spatially distinct populations of mesenchyme in the autopod. Finally, we perform scRNA-seq on murine embryonic limbs to facilitate cross-species developmental comparison at single-cell resolution, finding substantial homology between the two species.

developmental biology↗

bioRxiv2022

Prefusion-Stabilized Lassa Virus Trimer Identifies Neutralizing Nanobodies and Reveals an Apex-Situated Site of Vulnerability

Lassa virus (LASV) is responsible for 100,000-300,000 zoonotic infections annually and poses a threat to public health. Development of antibody-based therapeutics or vaccines has been challenging because neutralizing antibodies - even among Lassa hemorrhagic fever survivors - are generally of low titer, and the target of neutralizing antibodies, the trimeric glycoprotein complex (GPC), a type 1-fusion machine with GP1 and GP2 subunits, has been difficult to produce. Here, we use structure-based design to obtain a soluble LASV GPC by engineering an inter-protomer disulfide (R207GCGP1-L326CGP2) and appending the T4-fibritin trimerization domain. We verified the antigenicity of this prefusion-stabilized LASV GPC against a panel of human antibodies and used electron microscopy (EM) to confirm its trimeric association. We panned the prefusion-stabilized LASV GPC against single domain nanobody libraries and identified one of camel origin, which we named D5, which bound GPC with 27 nM affinity and neutralized the Josiah strain of LASV with an IC50 of 12 {micro}g/ml when formatted into a bivalent IgG2a context. The cryo-EM structure of a ternary complex of the D5 nanobody, the antigen-binding fragment of human antibody 8.11G, and LASV GPC revealed D5 to recognize a site-of-vulnerability at the trimer apex. The recognized site appeared to be specific to GPC lacking cleavage of between GP1 and GP2 subunits. Collectively, our findings suggest that GPC-cleavage intermediates may be targets for LASV neutralization and define an apex-situated site of vulnerability for vaccine development. SignificanceLassa virus (LASV) infection is expanding outside its traditionally endemic areas in West Africa, posing a biothreat to the world. LASV-neutralizing antibodies, moreover, have proven difficult to elicit. To gain insight into requirements for antibody-mediated neutralization of LASV, we developed a prefusion-stabilized LASV glycoprotein trimer (GPC), panned it against phage libraries comprised single-domain antibodies or nanobodies from shark and camel, and identified one, D5, which - when placed into bivalent IgG2a context - could neutralize LASV. Cryo-EM analysis revealed D5 to recognize a cleavage-dependent site-of-vulnerability at the trimer apex. We propose this apex-situated site to be an attractive target for LASV vaccine and therapeutic development.

immunology↗

bioRxiv2022

In Silico Improvement of Highly Protective Anti-Malarial Antibodies

Antibody CIS43 binds Plasmodium falciparum circumsporozoite protein (PfCSP) and protects against malaria, as recently demonstrated clinically. To improve the efficacy of CIS43, we developed an in silico pipeline to optimize the interaction energy of CIS43 to its junctional epitope (peptide 21: PfCSP residues 101-115). Starting from two improved CIS43 variants, recently elicited from a CIS43-germline knock-in mice, single and double amino acid substitutions in the peptide 21-proximal heavy (VH) and light (VL) variable regions were introduced. CIS43-variants, selected on the basis of improved in silico interface and stability energies, showed increased affinity to peptide 21 and superior malaria-protective efficacy. The best designed variant, antibody P3-43, was significantly more protective than its template antibody m43.151, with greater liver-burden protection than the current best-in-class (antibody iGL-CIS43.D3). Crystal structures of improved antibodies revealed atomic-level interactions explaining gains in binding affinity. The reported pipeline provides a powerful in silico approach to improve antibody functionality.

immunology↗

bioRxiv2022

Model guided design of enhanced bi-stable controllers to effectively switch cellular states

Bi-stable gene regulatory motifs are found in a wide variety of natural gene regulatory networks and effect transcriptional switching between stable phenotypic states in cells. In synthetic gene regulatory circuits, these architectures can be leveraged to dynamically switch between distinct metabolic states for metabolic engineering and therapeutic applications. However, the lack of modularity and predictability of these motifs in varying environments has limited widespread application, especially since the factors that affect switching characteristics are still unclear. Using a genome-scale metabolic model, we first show that the productivity of a two-stage bioprocess declines as switching is delayed, establishing the value of fast-switching motifs. We then use a mathematical model, along with a newly developed dynamical modeling and continuity analysis framework, to analyze the dynamics and robustness of bi-stable switches over a range of biologically relevant parameter values, and identify a trade-off between the robustness of the motif - the parameter range over which it retains bi-stable function - and the speed at which it effects a phenotypic change. Crucially, by extending our model to incorporate host growth rate through a proteome-partitioning framework, we find that growth rate reshapes the achievable speed-robustness frontier, with the most robust switches attainable at an intermediate growth rate, and that the optimal balance of repressor degradation rates is set by both their relative binding affinities and the growth rate of the host. Using E. coli as a model host, we then constructed a library of 100 transcriptional switches spanning a wide range of switching speeds, experimentally confirming this trade-off and revealing host growth rate as a third design axis - alongside robustness and switching speed - that shapes switch behavior. Together, these results establish design principles for building bi-stable switches with defined switching characteristics, and we anticipate that our library of experimentally validated bi-stable switches will be valuable for effecting phenotypic changes with differing switching-speed requirements in metabolic engineering applications.

synthetic biology↗

bioRxiv2022

Convergent impact of schizophrenia risk genes

Genetic studies of schizophrenia reveal a complex polygenic risk architecture comprised of hundreds of risk variants; most are common in the population at-large, non-coding, and act by genetically regulating the expression of one or more gene targets ("eGenes"). It remains unclear how genetic variants predicted to confer individually small effects combine to yield substantial clinical impacts in aggregate. Here, we demonstrate that eGenes have shared downstream transcriptomic effects ("convergence") that may underlie unexpected interactions ("non-additive effects") observed when eGenes are manipulated in combination. We apply a pooled CRISPR approach to perturb schizophrenia eGenes in human induced pluripotent stem cell-derived glutamatergic neurons. The strength and specificity of convergence increased between functionally similar eGenes. Predicting that convergence might impact additive relationships between risk loci when inherited together, we use an arrayed approach to explore bidirectional combinatorial perturbations of a partially overlapping set of fifteen schizophrenia eGenes. When specifically considering groups of synaptic or epigenetic eGenes, combinatorial eGene perturbations yield changes that are smaller than predicted by summing individual eGene effects ("sub-additive effects"). Moreover, convergent and non-additive downstream transcriptomic effects overlap, suggesting that functional redundancy of eGenes may be a major mechanism underlying non-additivity. Combinatorial perturbations result in outcomes that are not yet well-predicted by single eGene perturbations alone, indicating that the effects of polygenic risk cannot necessarily be extrapolated from experiments testing one risk gene at a time.

neuroscience↗

bioRxiv2022

PPML-Omics: a Privacy-Preserving federated Machine Learning system protects patients' privacy from omic data

Modern machine learning models towards various tasks with omic data analysis give rise to threats of privacy leakage of patients involved in those datasets. Despite the advances in different privacy technologies, existing methods tend to introduce too much computational cost (e.g. cryptographic methods) or noise (e.g. differential privacy), which hampers either model usefulness or accuracy in protecting privacy in biological data. Here, we proposed a secure and privacy-preserving machine learning method (PPML-Omics) by designing a decentralized version of the differential private federated learning algorithm. We applied PPML-Omics to analyze data from three sequencing technologies, and addressed the privacy concern in three major tasks of omic data, namely cancer classification with bulk RNA-seq, clustering with single-cell RNA-seq, and the integration of spatial gene expression and tumour morphology with spatial transcriptomics, under three representative deep learning models. We also examined privacy breaches in depth through privacy attack experiments and demonstrated that PPML-Omics could protect patients privacy. In each of these applications, PPML-Omics was able to outperform methods of comparison under the same level of privacy guarantee, demonstrating the versatility of the method in simultaneously balancing the privacy-preserving capability and utility in practical omic data analysis. Furthermore, we gave the theoretical proof of the privacy-preserving capability of PPML-Omics, suggesting the first mathematically guaranteed method with robust and generalizable empirical performance in protecting patients privacy in omic data.

bioinformatics↗